Muscle, joint & pain

Tennis Elbow: Why the Best Treatment Is Often Time

Save

Most people arrive at this question already holding a recommendation for a cortisone shot. The trial that tested that shot against a placebo injection found the shot came out behind at twelve months, with more relapses. This page lays out what that means, why an injection gets offered anyway, where injections and surgery genuinely do earn their place in other conditions, and how to interrogate the next PRP claim you are shown.

Last updated: July 2026

Talk to a clinician

Gale can help you find a clinician in your state and request a visit.

Find care →

What did the best tennis elbow trial actually find?

It found roughly the opposite of what everyone expects, in both directions at once. In a randomized controlled trial of people with unilateral lateral epicondylalgia, corticosteroid injection produced worse outcomes at one year, and higher recurrence, than a placebo injection. In the same trial, physiotherapy delivered no significant added benefit at one year 1.

Cortisone was worse than a placebo injection at one year, and physiotherapy added nothing significant at one year, in the same trial 1.

Sit with both halves, because most write-ups keep only the half they like. The surgical and injection world tends to quote the physiotherapy result; the rehabilitation world tends to quote the cortisone result. The trial said both. The shot people are most often handed did measurable harm across the year compared with a dummy shot, and the exercise programme people are most often prescribed did not measurably change where they ended up twelve months later.

That is an unusually clean piece of evidence for a condition this common, and it reframes the question in the title. "Injection or surgery, what works" assumes the contest is between active treatments. In tennis elbow, the more honest contest is between any active treatment and the passage of time.

Why is time a real answer here and not a shrug?

Because of what happens when you hold the two findings side by side. Across twelve months, the injection did worse than a dummy injection, and the exercise programme did not significantly alter the destination 1. Something carried those people through the year, and on the trial's own evidence it was not the treatments being tested. That is what "time" means in this context. It is an inference, and it is worth naming as one, but it is the inference the numbers support.

Slow unaided recovery is an ordinary event in this field rather than a consolation prize invented for the elbow. Frozen shoulder moves through freezing, frozen, and thawing stages and usually resolves over one to three years, with range-of-motion physical therapy as the primary treatment 2. A lumbar herniated disk pressing on a nerve root causes sciatica, and most people improve within weeks to months without surgery, with only a small percentage ever needing an operation 3.

Frozen shoulder, a different condition entirely, usually resolves over one to three years on its own course 2.

Those are other conditions, cited for one narrow purpose: to establish that a musculoskeletal problem can take an infuriatingly long time and still end well without anyone intervening. Which is exactly why treatment claims in this territory need reading with care. When a condition improves on its own over a year, anything administered at the worst moment will appear to have caused the improvement. That is regression to the mean, and it is the whole reason placebo-controlled trials like this one have to exist. Tennis elbow healing time is the variable that makes almost every uncontrolled claim about this condition look impressive.

So why does everyone keep offering me an injection?

Because low-value care is a structural habit rather than an individual failing, and the pressures that produce it are the same everywhere. A Lancet call for action documented that low-value care for low back pain — unnecessary imaging, opioids, injections, and surgery — is widespread globally and ought to be reduced 4. That paper is about backs rather than elbows, and the honest caveat is that it does not study your joint. But the machinery it describes is not anatomical.

The rest is human. An injection is fast. It is concrete. It is something done rather than something endured, and by the time a person reaches the appointment they have usually spent months enduring. The clinician wants to help, the patient wants relief, and a needle satisfies both wishes in ten minutes. Nothing about that requires bad faith from anyone in the room.

Which is why the tennis elbow cortisone conversation is worth having explicitly rather than by default. If a shot is offered, the useful question is not whether it will help this month. It is what the trial evidence says about where you stand a year from now, and whether the person offering it knows that evidence exists.

Injections are not universally wrong, and surgery is not universally low-value

This is the point where a page like this can do real damage if it overreaches, so read the boundary carefully. A finding about corticosteroid in tennis elbow is a finding about corticosteroid in tennis elbow. It does not travel to every drug, every tendon, or every joint. For carpal tunnel syndrome, the orthopaedic guideline places splinting and corticosteroid injection within nonsurgical management, and surgical release for appropriate patients 5. Same drug, different condition, opposite recommendation.

The cortisone finding belongs to tennis elbow. Same drug, different condition, different answer 15.

Surgery, likewise, genuinely wins some head-to-head contests. In the UK FASHIoN trial, hip arthroscopy for femoroacetabular impingement syndrome produced modestly better patient-reported hip function at twelve months than personalised physiotherapist-led conservative care, albeit at substantially higher cost 6. That is a randomized comparison in which the operation came out ahead of good conservative treatment. Whoever tells you that surgery is reliably the low-value choice is selling a story just as lazy as the one that says a scan finding must be cut out.

The defensible position is narrower and less quotable than either camp's. Each condition has its own evidence, the answers do not rhyme, and the work is finding out what your condition's evidence says rather than importing a slogan from somebody else's joint. The rotator cuff repair decision, the hip, the spine and the elbow all have different answers, and they were all arrived at the same way: by testing.

What about PRP, and what about surgery on the elbow itself?

This page cannot tell you, and it is going to say so rather than improvise. The sources it is built on do not test platelet-rich plasma for tennis elbow, and they do not test tennis elbow surgery. Writing a confident paragraph about either would mean manufacturing evidence, and a stated gap is worth more to you than a fabricated answer.

What is transferable is the method for interrogating the claim when somebody makes it to you. The trial above is a usable template, and these are the questions it implies:

  • What was the comparison group? Against a placebo injection, or against nothing at all? A treatment that beats nothing has beaten very little, because sticking a needle in someone is itself an event.
  • How long was the follow-up? Twelve weeks or twelve months? Tennis elbow punishes short follow-up more than almost anything, and the trial reported not just worse one-year outcomes but higher recurrence 1 — which is the shape of something that helps and then unwinds.
  • What outcome was counted? Pain sitting still, or the capacity to carry a full kettle without dread?
  • Who paid, and who else has replicated it? A single enthusiastic series is a hypothesis, not a result.

One practical thing you can do meanwhile: there is a scored questionnaire built for this condition, the patient-rated tennis elbow evaluation, and completing it today and again in three months gives you a slope rather than a memory. Memory is a terrible instrument for a condition that fluctuates weekly.

When does escalating become the right call?

Honesty first: this page cannot hand you an evidence-based threshold for tennis elbow surgery, because no source behind it contains one, and inventing a threshold would be exactly the failure it has spent six sections warning about. What it can do is name the things that legitimately change the conversation, each of which is a question to raise out loud rather than a rule to apply to yourself.

  • Whether the diagnosis is still the diagnosis. Months of failure is a reason to re-examine the label, not just to escalate the treatment of it. Whether this is tennis vs golfer's elbow, sitting on opposite sides of the joint, is worth settling before anyone treats either. Nerve problems and referred pain can also wear a tendon's clothing.
  • Whether real time has genuinely passed. The anchoring trial followed people for a full year 1. A decision made at week six is a decision made before the evidence's own clock has finished running.
  • What function is actually doing. A recorded score across months, not an impression. A slope that is flat and liveable is a different situation from one still heading down.
  • What the specific operation claims to fix, and on what evidence. Ask the surgeon directly what the trial data are for the procedure in someone with your presentation. If the answer is thin, that is information too.

None of this argues that operating on an elbow is wrong. It argues that the reason to operate ought to be better than accumulated frustration, and that frustration is precisely what a slow natural history manufactures. The sequence here is unglamorous: get the diagnosis right, give the condition the time its own evidence assumes, measure honestly, and escalate from information rather than from exhaustion.

Common questions

In the randomized trial that tested it, corticosteroid injection produced worse outcomes at one year and higher recurrence than a placebo injection. That is a finding about the twelve-month horizon rather than about how the arm feels next week. It is worth raising explicitly with whoever offers the shot, and worth asking whether they are weighing that year-out evidence against the short-term appeal.

This page will not answer that, because none of the sources behind it tested platelet-rich plasma, and a confident paragraph would be fabrication. What transfers is how to read the claim when someone makes it: ask what the comparison group received, how long people were followed, what outcome was counted, and whether anyone independent has reproduced the result.

Operations for this condition exist, but no source behind this page tested one, so it offers no verdict either way. The honest move is to ask the surgeon what randomized evidence supports the specific procedure in someone with your presentation, over what follow-up, against what comparison. A thin answer to that question is itself worth knowing before consenting.

Because the one-year endpoint describes where people finished, not the road they walked. A supervised loading programme is low-risk, it carries none of the recurrence signal the injection did, and it gives you a way to engage with a condition whose central cruelty is its duration. What the trial should end is the belief that skipping it costs you the final outcome.

No source behind this page gives a timeline for tennis elbow specifically, so it will not invent one. What the design hints at is the scale: the anchoring trial followed people for a full year, which is the horizon its investigators thought the condition needed. Other slow musculoskeletal conditions run on similar clocks, measured in many months rather than weeks.

Related

Say it back

How would you explain this to someone you love?

Two or three sentences, just as you’d say it. Gale reflects back what you focused on — a mirror, not a quiz.

Talk to a clinician

Gale can help you find a clinician in your state and request a visit.

Find care →

Elbow symptoms that are not a tendon problem

  • —Elbow pain after a fall or a direct blow with visible deformity, or a sudden inability to straighten or bend the elbow.
  • —An elbow that becomes hot, red, and swollen over hours, particularly alongside fever or chills.
  • —Numbness, pins and needles, or weakness in the hand — a weakening grip, or difficulty spreading the fingers — which points at a nerve rather than at a tendon.
  • —Elbow pain that wakes you regardless of how the arm is positioned, or elbow pain with unexplained weight loss or a history of cancer.

A hot, red, swollen elbow with fever needs to be seen the same day, and an obviously deformed elbow after a fall belongs in an emergency department. Call 911 if the hand or arm becomes numb, cold, or pale.

Gale's health library explains what the published evidence does and does not show. It is not medical advice, it has not examined your elbow, and it is not a substitute for a clinician who can.

References

  1. 1.Coombes BK, Bisset L, Brooks P, Khan A, Vicenzino B (2013). Effect of Corticosteroid Injection, Physiotherapy, or Both on Clinical Outcomes in Patients With Unilateral Lateral Epicondylalgia: A Randomized Controlled Trial. JAMA. PMID 23385272 ✓For lateral epicondylalgia, corticosteroid injection produced worse one-year outcomes and higher recurrence than placebo injection, and physiotherapy gave no significant added benefit at one year — the central evidence for the article, including the one-year follow-up horizon and the recurrence signal.
  2. 2.American Academy of Orthopaedic Surgeons (OrthoInfo) (2024). Frozen Shoulder (Adhesive Capsulitis). OrthoInfo — AAOS. link ✓Frozen shoulder progresses through freezing, frozen, and thawing stages and usually resolves over one to three years, with range-of-motion physical therapy as the primary treatment — cited as a separate condition establishing that slow unaided resolution is ordinary in musculoskeletal care.
  3. 3.American Academy of Orthopaedic Surgeons (OrthoInfo) (2024). Herniated Disk in the Lower Back. OrthoInfo — AAOS. link ✓A lumbar herniated disk can compress a nerve root causing sciatica; most people improve within weeks to months without surgery, and only a small percentage require an operation — cited as a separate condition illustrating favourable natural history.
  4. 4.Buchbinder R, van Tulder M, Öberg B, et al. (2018). Low back pain: a call for action. The Lancet. doi:10.1016/S0140-6736(18)30488-4Low-value care for low back pain — unnecessary imaging, opioids, injections, and surgery — is widespread globally and should be reduced; cited only for the low-value-care phenomenon, with its low-back-pain scope stated explicitly in the text.
  5. 5.American Academy of Orthopaedic Surgeons (AAOS) (2016). Management of Carpal Tunnel Syndrome — Evidence-Based Clinical Practice Guideline. Journal of the American Academy of Orthopaedic Surgeons. doi:10.5435/JAAOS-D-17-00451 ✓For carpal tunnel syndrome the orthopaedic guideline includes splinting and corticosteroid injection within nonsurgical management, and surgical release for appropriate patients — cited to show that the tennis elbow corticosteroid finding does not generalise across conditions.
  6. 6.Griffin DR, Dickenson EJ, Wall PDH, et al. (UK FASHIoN) (2018). Hip arthroscopy versus best conservative care for the treatment of femoroacetabular impingement syndrome (UK FASHIoN): a multicentre randomised controlled trial. The Lancet. doi:10.1016/S0140-6736(18)31202-9For femoroacetabular impingement syndrome, hip arthroscopy produced modestly better patient-reported hip function at twelve months than personalised physiotherapist-led conservative care, at substantially higher cost — cited as a randomized comparison in which surgery outperformed good conservative care.

6 sources, numbered by first appearance. General health information, not medical advice. AI-assisted editorial content — citations link their sources. Editorial policy

← Back to the health library